English

• Photocatalytic CO₂ reduction reaction (CO₂RR) is one of the most promising and green methods of CO₂ conversion and utilization [1-4] for the reason that it can exploit solar energy as a driving force to convert low-value CO₂ into high-value carbon-based chemicals like CH₄, CO, CH₃OH, etc. [5-9]. At present, molecular metal compounds usually exhibit excellent photocatalytic performance such as high turnover number (TON) and turnover frequency (TOF) among the reported CO₂-photoreduction catalysts [10-12] due to the high active site utilization rate and the essence of homogeneous catalysis [13-16]. Particularly, the vast majority of molecular metal compounds can provide a clear single-crystal structure [17-19], allowing the catalytic active sites to be identified accurately and studying the reaction mechanism [20-22]. Based on these advantages [23], molecular metal compounds have the great potential to be used as model catalyst systems to understand the corresponding structure-property relationship [24-27], and thus guide for designing and synthesizing high-efficiency CO₂RR catalysts [28, 29].

  Generally, compared to monometallic catalysts, bimetallic catalysts always tend to show better performance in many catalytic reactions because of the synergistic effect of two interacted metal sites [30-34]. Unfortunately, only few suitable model systems were reported to specifically investigate the synergistic effect of two adjacent metal sites by means of second metal (activity or inactive) introduction on photocatalytic CO₂RR performance [35, 36]. In many cases, metal ion-doping is the most common strategy used for surveying the influence of synergistic effect of two metal sites on the catalytic performance [37-39]. Although the ratio of two metal ions can be effectively adjusted by this method [40, 41], it is hard to accurately control the position and coordination environment of the two specific metal ions [42, 43], which affects the precise evaluation for the structure-property relationship [37]. In order to overcome this issue, the best way is to design and synthesize a well-defined model catalyst system that can accurately identify each catalytic site [44]. In the previous reports, only one molecular metal compound catalyst investigated the effect of the introduction of an inactive metal ion nearby active metal site on photocatalytic CO₂RR performance, but there are no direct interactions between the two involved metal sites [35]. If there are direct bonding interactions existed between two metal sites, the introduction of the secondary inactive metal site may significantly change the molecular orbitals and charge transfer characteristics of the active metal site, and thus affecting the final catalytic performance [45].

  In light of these thoughts, a molecular metal compound model catalyst system including mononuclear [NiL¹(CH₃OH)] (labeled as Ni-L¹), [CuL¹(H₂O)] (labeled as Cu-L¹), dinuclear homometallic [Ni₂L²(CH₃OH)₂] and [Cu₂L²(CH₃OH)] (labeled as Ni₂-L² and Cu₂-L²), and dinuclear heterometallic [CuNiL²] (labeled as CuNi-L²) was designed and synthesized. Single crystal X-ray diffraction result shows that the Cu-L¹ and Ni-L¹ are isostructural, Ni₂-L², Cu₂-L² and CuNi-L² are isostructural. Importantly, these five molecular metal compounds can be regarded as a model catalyst system to systematically explore the influences of active monometallic site, bimetallic sites and heterometallic sites on the photocatalytic performance of CO₂RR in homogeneous conditions. The photocatalytic CO₂RR measurement results showed that both Ni-L¹ and Ni₂-L² exhibited high photocatalytic CO₂-to-CO activity, but neither Cu-L¹ nor Cu₂-L² showed catalytic activity, indicating that Ni^II is the photocatalytic active center while Cu^II is the inactive site. Moreover, under 8 h of visible light irradiation and the same concentration of Ni^II (0.8 µmol/L), the photocatalytic performance of Ni₂-L² (2.18 µmol) is better than that of Ni-L¹ (1.52 µmol), manifesting the dual active Ni^II sites show more advantageous catalytic activity owing to the synergistic effect. It is worth noting that for dinuclear heterometallic CuNi-L², the single active Ni^II site is coupled with inactive Cu^II ion, under the same photocatalytic conditions as Ni-L¹ and Ni₂-L², the CO yield of the photoreduced product is calculated to be 4.60 µmol (CO selectivity 93.5%), which is 2.1 and 3.0 times higher than that of Ni₂-L² and Ni-L¹, respectively. It reveals that the introduction of the secondary inactive Cu^II site can further promote the photocatalytic performance of the adjacent interacted active Ni^II ion. The density functional theory (DFT) calculation results declared that the electron cloud distribution around catalytic active site Ni^II ion is effectively enhanced owing to the introduction of relatively catalytic inactive Cu^II ion, which increases the electron-acquiring ability of active Ni^II site and significantly accelerates the photocatalytic CO₂RR. More importantly, this work represents a well-defined model catalyst system to effectively reveal the important influence of the bimetallic synergy between active and inactive metal ions on the photocatalytic CO₂RR performance by adjusting the species, number and location of metal active centers in catalysts.

  The CuNi-L² was synthesized through the precursor CuNa₂-L². A 50 mL EtOH solution of copper acetate (2.00 g, 0.01 mol) was slowly added to a 50 mL H₂O solution of H₄L² (3.84 g, 0.01 mol) and NaOH (0.8 g, 0.02 mol) [46, 47]. The mixture was heated under reflux under N₂ atmosphere for 30 min followed by concentrating the resulting mixture to give a brown product CuNa₂-L² with quantitative yield. After acidification and recrystallization, crystal CuH₂-L² was obtained which further proved CuNa₂-L² synthesized successfully. Ni(ClO₄)₂·6H₂O (50 mg) and CuNa₂-L² (10 mg) was dissolved in 8 mL N, N-dimethylformamide (DMF) ultrasonically, then 2 mL EtOH was added into the solution. This solution was transferred into a 10 mL glass vial for 72 h at 353 K under autogenous pressure. After cooling down to room temperature, brown crystals were collected by filtration and fully washed with MeCN. The crystals were soaked in 20 mL of CH₃CN for 2 days replacing the solvent every 12 h. Finally, filtered and dried under vacuum. See supporting information for the specific synthesis methods of the other compounds.

  Photocatalytic reduction of CO₂ was performed in a 50 mL quartz reactor with as-prepared crystals. Photocatalyst (0.02 µmol, Ni-L¹, Ni₂-L² and CuNi-L² based on the Ni^II site, Cu-L¹ and Cu₂-L² based on the Cu^II site), [Ru(bpy)₃]Cl₂·6H₂O (bpy = 2,2′-bipyridine, 0.02 mmol) were added into the mixed solution which contained CH₃CN (20 mL), H₂O (5 mL) and triisopropanolamine (TIPA, 5 mL) as an electron donor. After degassing with high-purity CO₂ to remove the dissolved O₂ for 30 min, the reaction was performed under the irradiation of a 300 W Xe lamp with UV-cut to keep the wavelengths in the range of exceeding 420 nm. The reaction temperature was kept at 298 K by using the cooling water circulation. In order to detect the content of reduction product produced by the reaction mixture, 500 µL of gas-product was extracted from the reactor with a syringe and injected into the GC equipped a flame ionization detector (FID) with a methanizer and a thermal conductivity detector (TCD), using nitrogen as the carrier gas. By comparing the integrated area of the gas-phase product with the calibration curve, the volume of carbon monoxide and hydrogen can be calculated. Each photocatalytic reaction was repeated at least three times to confirm the reliability of the data.

  Single crystal X-ray diffraction result shows that the mononuclear Ni-L¹ crystallizes in triclinic system with the P-1 space group (Table S1 in Supporting information), and the asymmetric unit contains a transition metal ion Ni^II, a deprotonated Schiff base ligand (L¹)²⁻ and a free methanol molecule. The only crystallographic Ni^II center is captured by the N₂O₂ chelating pocket (two N atoms from two C=N bonds and two O atoms from two phenolic hydroxyl groups) to form a planar quadrilateral coordination configuration. It is noteworthy that the Ni^II ion and its coordination atoms in Ni-L¹ are totally in the same plane. This means that there is sufficient coordination space in axial position of the metal ion to allow small molecules to attack. Moreover, the Ni^II ion in Ni-L¹ can be replaced by Cu^II ion, namely the isostructural Cu-L¹, which crystallizes in the orthorhombic system with a Pnma space group (Table S1). In addition, we further introduced carboxyl groups in the ortho positions of the two phenolic hydroxyl groups of H₂L¹ to synthesize the H₄L² ligand containing two metal ion chelating pockets (N₂O₂ and O₄ coordination modes). The ligand can catch hold of two identical or different metal ions effectively. Furthermore, just like the original N₂O₂ cavity, all coordination atoms of the constructed parallel secondary O₄ coordination mode are also composed of a planar quadrilateral geometry. When the H₄L² ligand captures two Ni^II ions, a dinuclear homometallic compound Ni₂-L² is synthesized and it crystallizes in triclinic system with the space group P-1 (Table S2 in Supporting information). Ni1 atom is located in N₂O₂ chelating pocket, while Ni2 atom is captured by another O₄ chelating pocket. Ni1 and Ni2 atoms are connected and interacted with each other by sharing two phenolic hydroxyl O atoms. Similarly, the H₄L² ligand can also clamp two Cu^II ions to form an isostructural homometallic Cu₂-L², but which crystallizes in monoclinic system with the P2₁/c space group (Table S2). More importantly, we can fix Cu^II and Ni^II ions simultaneously by these two kinds of different chelating pockets of H₄L² ligand though stepwise synthesis method to construct isostructural dinuclear heterometallic compound CuNi-L², which crystallizes in monoclinic system with C2/c space group (Table S2). For CuNi-L², Cu1 atom is trapped into the N₂O₂ chelating pocket, and Ni1 atom is fastened by the O₄ coordination mode. It is worth noting that due to the similar configuration of these compounds, this series of compounds Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² (Fig. 1) can serve as model systems to study the effects of monometallic, bimetallic (homometallic and heterometallic) compounds on the catalytic performance for specific small molecule activity. More importantly, the performance impact produced by the synergistic effect between dinuclear metal ions can be explored systematically.

  Figure 1

  

  Figure 1.  Structures of the molecular complexes. (a) mononuclear Ni-L¹; (b) mononuclear Cu-L¹; (c) dinuclear Ni₂-L²; (d) dinuclear Cu₂-L²; (e) dinuclear CuNi-L². All hydrogen atoms are omitted for clarity. Cu, yellow; Ni, green; O, red; N, blue; C, gray.

  DownLoad: Full-Size Img PowerPoint

  The phase purity of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² were confirmed by well-matched powder X-ray diffraction (PXRD) patterns (Figs. S1-S5 in Supporting information). According to the X-ray photoelectron spectroscopy (XPS) and inductively coupled plasma optical emission spectrometer (ICP-OES) results (Table S3 in Supporting information), the valence of Ni and Cu is + 2 (Fig. 2a and Fig. S6 in Supporting information) and the stoichiometric ratio of Cu/Ni in CuNi-L² is 1:1. Furthermore, the thermogravimetric analysis (TGA) curves of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² showed that their structures can be maintained well below 300 ℃ under air atmosphere (Fig. S7 in Supporting information). To evaluate the potential of the five compounds as photocatalysts, we performed a series of characterizations to assess their band structures. The highest occupied molecular orbital (HOMO) positions of the five molecular compounds were evaluated approximately by cyclic voltammetry (CV) (Fig. 2b, Figs. S13-S16 in Supporting information). Thus, the HOMO energy levels of the compounds were estimated to be 6.59 eV (1.74 V vs. NHE), 6.73 eV (1.88 V vs. NHE), 6.63 eV (1.78 V vs. NHE), 6.77 eV (1.92 V vs. NHE), 6.40 eV (1.55 V vs. NHE) for Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L², respectively. The bandgaps (E_g) of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² were estimated by the UV–vis absorption spectra (Fig. 2c). The E values were calculated to be 2.36, 2.51, 2.46, 2.51 and 2.17 eV, respectively. Then the lowest unoccupied molecular orbital (LUMO) positions of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² were calculated to be −0.62 V, −0.63 V, −0.68 V, −0.59 V, −0.62 V vs. NHE, pH 7. As the LUMO positions of these compounds are more negative than the redox potentials of most photocatalytic CO₂RR products such as HCOOH (−0.58 V vs. NHE, pH 7), CO (−0.51 V vs. NHE, pH 7) or CH₄ (−0.24 V vs. NHE, pH 7) (Fig. 2d), these compounds have potential for CO₂RR as catalysts [48].

  Figure 2

  

  Figure 2.  (a) High resolution Ni 2p XPS spectra of Ni-L¹, Ni₂-L² and CuNi-L² in survey scan. (b) Cyclic voltammogram of CuNi-L² in acetonitrile containing 0.1 mol/L Bu₄NPF₆ at a scan rate of 50 mV/s. (c) UV–vis absorption spectra of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² in the solution state (chloroform). (d) Band structure for Ni-L¹, Ni₂-L² and CuNi-L².

  DownLoad: Full-Size Img PowerPoint

  Taking all the above features into consideration, the photocatalytic CO₂RR of these compounds as homogeneous catalysts were tested under a pure atmosphere of CO₂ (1.0 atm, 25 ℃) in a 30 mL CH₃CN/H₂O solution (v/v = 4:1) with a certain amount (0.8 µmol/L) of catalysts, triisopropanolamine (TIPA) as electron donor and [Ru(bpy)₃]Cl₂·6H₂O as auxiliary photosensitizer (PS) under visible light irradiation (λ ≥ 420 nm) [11, 49]. After being irradiated, the generated gasses products were analyzed by gas chromatography (GC) (Fig. S17 in Supporting information) and the liquid products were tested by NMR. The results showed that the main gaseous reductive products of these compounds were CO and H₂, and no liquid reduction products were detected in the reaction system. As the irradiation time increased, the reduction products decreased gradually per unit time, and there is almost no product generating after 8 h. After one single photoredox cycle, 1.52 µmol of CO and 0.20 µmol of H₂ were produced in accordance with a CO selectivity of 88.4% and a TON_CO of 76 catalyzed by mononuclear compound Ni-L¹ (0.8 µmol/L) within 8 h. In contrast, as the catalyst replaced by another mononuclear compound Cu-L¹, almost no reduction products were detected. This consequence implied that the metal ion Cu^II is inactive in this catalytic system. Specifically, when the photocatalyst changed from the mononuclear complex Ni-L¹ to the bimetallic complex Ni₂-L² (0.4 µmol/L) with the same Ni^II ion concentration and irradiation time, the yield of CO was increased to 2.18 µmol (TON = 109), along with the selectivity of CO was also improved from 88.4% to 93.9% (Fig. 3a). The results showed that both of the Ni^II ions in the two types of coordination environment (N₂O₂ and O₄) have effective photocatalytic CO₂RR capabilities. There is no doubt that introducing the second metal has a certain positive effect on the product selectivity. However, the isomorphic compound Cu₂-L² produced only trace amounts of gaseous products, which confirmed that Cu^II ion expresses negligible catalytic activity in this photocatalytic system one step closer. In other words, Cu^II ion did not show an effective photocatalytic CO₂ reduction ability whether it was in the N₂O₂ coordination environment or the O₄ coordination environment. Interestingly, when CuNi-L² (0.8 µmol/L) acted as the catalyst, it showed the highest CO yield (4.60 µmol) among the nickel-containing catalysts (Ni-L¹, Ni₂-L² and CuNi-L²) under the identical photocatalytic condition. The TON_CO was up to 230, which was 2.1 and 3.0 times as much as that of Ni₂-L² and Ni-L¹ respectively. Meanwhile, the selectivity for CO was 93.5%, which was comparable to that of Ni₂-L². A summary of the specific selectivities, TONs and TOFs of these photocatalytic systems were provided in Table S4 (Supporting information).

  Figure 3

  

  Figure 3.  The photocatalytic CO₂RR was carried out under the conditions of a 30 mL CO₂-saturated CH₃CN/H₂O solution (v/v = 4:1) at 25 ℃ with the same concentration of Ni^II (0.8 µmol/L), TIPA (0.8 mol/L) and [Ru(bpy)₃]Cl₂·6H₂O (0.85 mmol/L) under visible light irradiation (λ ≥ 420 nm, irradiation area, 7.06 cm²). (a) Photocatalytic production of CO and H₂ catalyzed by Ni-L¹ (0.8 µmol/L), Ni₂-L² (0.4 µmol/L) and CuNi-L² (0.8 µmol/L). (b) Mounts of CO and H₂ produced as a function of the visible-light irradiation time over CuNi-L² (0.8 µmol/L). (c) Durability test for CuNi-L² (0.8 µmol/L). (d) The mass spectra of ¹³CO recorded under a ¹³CO₂ atmosphere.

  DownLoad: Full-Size Img PowerPoint

  Considering that CuNi-L² presented the highest photocatalytic activity and superior selectivity for CO₂-to-CO conversion, a collection of control tests with CuNi-L² as the representative example were carried out to identify the roles of catalyst and other influence factors in photocatalytic CO₂RR. The corresponding results were summarized in Table S4. No gas phase product was detected when the photocatalytic reaction was performed without CO₂ (replaced by N₂) (Table S4, entry 4), which indicated the detected CO originated from CO₂. The absence of gas production in the dark indicated that it is the truly photoexcited states to initiate the reaction (Table S4, entry 5). Moreover, the reaction system in the absence of TIPA exhibited trace evolution amount of CO, suggesting that the photoexcited states of CuNi-L² were reductively quenched by TIPA (Table S4, entry 6). When the photocatalytic system without CuNi-L² or photosensitizer, a trace amount of CO was detected (Table S4, entries 7 and 8). The above experimental results showed that the catalyst, photosensitizer, visible light, CO₂ atmosphere and TIPA are all the indispensable reaction conditions in this photocatalytic CO₂RR model system. At the same time, noting that CO production was also very low in the absence of H₂O (Table S4, entry 9), we conducted a series of experiments to explore the influence of different proportions of H₂O on CO₂RR. When the ratio of H₂O/CH₃CN was slightly increased to 1:9, the yield of CO (3.7 µmol) was increased significantly, indicating that H₂O acted as an important proton source in the reaction process. When the H₂O/CH₃CN ratio further increased to 1:4, the amount of CO was increased to 4.6 µmol, and the selectivity was also risen from 77.1% to 91.3% at the same time. However, as the H₂O/CH₃CN ratio reached to 1:1, the yield of CO decreased instead, and only a few CO produced in pure H₂O (Table S5 in Supporting information).

  After an 8-h photocatalysis, the total amount of gaseous products produced hardly increased (Fig. 3b), which may attribute to the inactivation of the photosensitizer. In order to confirm that, recycling experiments were carried out by taking CuNi-L² as an example. When almost no CO was produced after being illuminated for 8 h, the equal amounts of fresh PS was added into the system and pure CO₂ gas was reintroduced. As expected, the stagnated reaction was activated again, which indicated that the termination of the system results from the deactivation of the photosensitizer rather than the poisoning of CuNi-L² catalyst. Moreover, CuNi-L² completed at least three cycles of 8 h reaction with a high total CO output (9.23 µmol) and fine selectivity (91.6%) to CO (Fig. 3c). Additionally, the apparent quantum efficiency (AQE) of CuNi-L², Ni₂-L² and Ni-L¹ were also measured under the monochromatic light of 420 nm (see Supporting information). The corresponding results showed that the ɸ_CO of CuNi-L² was calculated to be 0.018%, which was 2-fold higher than Ni₂-L² (0.0086%) and around 3-fold higher than Ni-L¹ (0.0060%). In order to confirm that the produced CO indeed comes from CO₂ reduction, isotopic tracing experiment was performed with representative CuNi-L² catalyst in ¹³CO₂ atmosphere. After irradiation under visible light for 4 h, gas chromatography-mass spectrometry (GC–MS) was used to analyze the reaction products. The detected peak of m/z 29 was assigned to ¹³CO, the peaks of m/z 13 and 16 belong to the debris of ¹³C and ¹⁶O (Fig. 3d). It is clear from the result that ¹³CO was converted from the ¹³CO₂ atmosphere, which indicated that CO is derived from the photocatalytic CO₂ reduction instead of the compounds decomposing (catalyst/PS/TIPA) in the catalytic system. At the same time, the ¹³C NMR spectrum indicated that there were no liquid reduction products detected in a series of reactions above (Fig. S20 in Supporting information) [50, 51]. As a result, this fact proved that the catalysts had good selectivity, and it also demonstrated that these compounds did have the CO₂RR activity. In addition, the steady-state photoluminescence (PL) spectroscopy of CuNi-L² was further carried out to provide insight into the excited states recombination behavior (Fig. S21 in Supporting information). It can be observed that the fluorescence intensity decreases gradually as the catalyst concentration increases, indicating that CuNi-L² has a good separation efficiency of photogenerated carriers. Based on the above photocatalytic tests, both the mononuclear compound Ni-L¹ and the dinuclear compound Ni₂-L² showed high CO₂RR performance, indicating that the Ni^II ions in the two coordination environments (N₂O₂ and O₄) possess effective catalytic activities. Moreover, Ni₂-L² exhibits better catalytic ability than Ni-L¹ under the same Ni^II ion concentration, we guess that it may be caused by the certain metal synergistic catalysis between the two Ni^II sites in different coordination environments. However, when the catalyst was replaced by the isomorphic Cu-L¹ and Cu₂-L² respectively, they did not show effective photocatalytic activity. That is to say, the Cu^II ions in the two coordination environments hardly express photocatalytic performance. It is worth noting CuNi-L² involves one active site Ni^II, but it shows the highest catalytic activity among these five catalysts. We considered that the inactive Cu^II ion probably exerts a certain effect on improving the photocatalytic activity of neighboring active Ni^II site.

  The experimental results demonstrated that the photocatalytic CO₂RR activities of the five monomolecular compounds gradually decreased by CuNi-L², Ni₂-L², Ni-L¹, Cu₂-L² and Cu-L¹ successively. DFT calculation was employed to get insight into the activity of the photocatalytic conversion of CO₂-to-CO. Then the Gibbs free energy (G) of the CO₂RR paths are compared to explore the differences in photocatalytic activity of the five catalysts (Fig. 4a). The CO₂-to-CO reduction pathway is primarily composed of three steps: CO₂ activation and hydrogenation (CO₂ → *COOH), dehydration (*COOH → *CO), and CO desorption (*CO → CO), where the rate-determining step is the first path. The energy change from CO₂ to *COOH on CuNi-L² and Ni₂-L² are the lowest (0.81 eV and 0.85 eV), Ni-L¹ (1.44 eV), Cu₂-L² (1.57 eV) and Cu-L¹ (1.97 eV), suggesting the highest activity of dual TM centers (CuNi-L² and Ni₂-L²) for CO₂ reduction. For the two investigated transitional metal ions, Cu^II sites apparently present an inactive behavior for the CO₂RR due to large energy increase for the formation of *COOH. Although the above results are consistent with the photocatalytic activity of the catalysts, the energy change from CO₂ to *COOH of CuNi-L² is not far different from Ni₂-L², while the CO production of CuNi-L² is about 1.5 times than that of Ni₂-L² in experiment. The dual TM sites were expected to stabilize the *COOH intermediate with a bridge structure, but the adsorbate returns to a single Ni^II site after geometry optimization (Fig. S22 in Supporting information). That is to say, the coordination environment of the Ni^II ions is crucial during the photocatalytic possess. In order to further explore the structure-activity relationship between CuNi-L² and Ni₂-L², time-dependent DFT calculations were performed to investigate the excitation properties of CuNi-L² and Ni₂-L² systems. Fig. 4b illustrates the electron contributions of the first excited state, which is basically contributed by HOMO-LUMO transition, the charge density difference between the ground state (S₀) and the first excited state (S₁). It is clear from the result above that the excited electron of Ni₂-L² is concentrated around two Ni^II ions and the coordinated N, O atoms, while the electron of CuNi-L² is mainly concentrated on the Ni^II ion and O atoms, no electron accumulates around the Cu^II ion and N atoms. In addition, the charge analysis reveals that the charge transfer from Ni^II ions to N₂O₂ and O₄ coordination are 0.21 and 0.47 respectively (Fig. 4c), indicating that Ni^II ions in the two coordination environments both have good reduction ability, of which the Ni^II in O₄ coordination possesses better reduction activity. Meanwhile, comparing the free energy change between the two Ni^II sites, the Ni^II in O₄-coordination expresses a lower energy change from CO₂ to *COOH than Ni^II in N₂O₂ coordination environment, which further proves the result above. The charge transfer in CuNi-L² for metal ions are 0.65 (Ni^II) and 0.01 (Cu^II) respectively. It can be found that the existence of inactive Cu^II ion promotes the accumulation of electrons on Ni^II, which causes the Ni^II with O₄ coordination environment in CuNi-L² expressing better reduction ability than that of Ni₂-L², and then significantly accelerates the photocatalytic reduction reaction of CO₂ to CO.

  Figure 4

  

  Figure 4.  (a) The free-energy profile for the CO₂RR pathway. (b) Charge density difference plot for the first excited state of CuNi-L² (left) and Ni₂-L² (right). (c) 2D display of charge density of CuNi-L² (left) and Ni₂-L² (right) systems. (d) Proposed photocatalytic mechanism of the molecular metal compounds for the CO₂ to CO conversion.

  DownLoad: Full-Size Img PowerPoint

  According to the above DFT calculation results, the corresponding photocatalytic CO₂RR mechanism can be proposed (Fig. 4d). Firstly, [Ru(bpy)₃]Cl₂·6H₂O is excited by the irradiation of visible light. Since the LUMO positions of the five compounds are lower than that of [Ru(bpy)₃]Cl₂·6H₂O, the photo-generated electrons can be transferred to the catalysts. At the same time, CO₂ molecules are adsorbed and activated by the exposed metal sites of the catalysts. Secondly, the adsorbed CO₂ on the active metal center obtains an electron and a proton to form a *COOH intermediate. Next, the *COOH intermediate gets an electron and a proton, then the *CO species is generated after dehydration from the *COOH. Finally, CO is generated by the desorption of *CO separating from the surface of the catalysts. The sacrificial electron donor was performed by TIPA to fill the holes and reductive quench the excited photosensitizer. The catalysts regenerated and enter the next cycle of CO₂-to-CO conversion.

  In summary, we design and synthesize five molecular metal compounds, containing mononuclear Ni-L¹ and Cu-L¹, dinuclear homometallic Ni₂-L², Cu₂-L² and dinuclear heterometallic CuNi-L². These molecular metal compounds can be regarded as a model catalyst system to systematically probe the influence on photocatalytic CO₂RR performance caused by monometallic site, bimetallic sites and heterometallic sites by precisely controlling of the species, number and catalytic properties of metal ion centers. In particular, it explores the direct effect of introducing inactive metal ions into bimetallic catalysts on the photocatalytic CO₂RR performance of two neighboring metal centers. When these molecular metal compounds act as homogeneous catalysts, the inactive Cu-introduced dinuclear heterometallic CuNi-L² (0.8 µmol/L) exhibits the highest photocatalytic selectivity (93.5%) and activity (4.60 µmol) after an 8-h photocatalysis, which is 2.1 and 3.0 times than that of dinuclear homometallic catalyst Ni₂-L² (2.18 µmol) and the mononuclear catalyst Ni-L¹ (1.52 µmol) respectively with the same concentration of Ni^II (0.8 µmol/L). The related DFT calculation results also show that, compared with Ni₂-L², the photo-generated electrons in CuNi-L² are promoted transferring to the adjacent active Ni^II site due to the introduction of inactive Cu^II ion. This fact improves the utilization efficiency of photo-generated electrons, and then the photocatalytic CO₂RR activity of Ni^II ions. Significantly, more insights were provided by this work for the future design and construction of high-efficiency bimetallic photocatalysts applied to photocatalytic CO₂RR.

  Declaration of competing interest

  The authors report no declarations of interest.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (NSFC, Nos. 92061101, 21871141, 21871142 and 22071109); Project funded by China Postdoctoral Science Foundation (No. 2018M630572).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.01.039.

  
  
• 1. [1]

     X. Chang, T. Wang, P. Yang, G. Zhang, J. Gong, Adv. Mater. 31 (2019) 1804710. doi: 10.1002/adma.201804710

  2. [2]

     Y. Hu, F. Zhan, Q. Wang, et al., J. Am. Chem. Soc. 142 (2020) 5618–5626. doi: 10.1021/jacs.9b12443

  3. [3]

     M. Sun, C. Wang, C.Y. Sun, et al., J Catal. 385 (2020) 70–75. doi: 10.1016/j.jcat.2020.03.005

  4. [4]

     S. Navarro-Jaén, M. Virginie, J. Bonin, et al., Nat. Rev. Chem. 5 (2021) 564–579. doi: 10.1038/s41570-021-00289-y

  5. [5]

     W. Tu, Y. Zhou and Z. Zou, Adv. Mater. 26 (2014) 4607–4626. doi: 10.1002/adma.201400087

  6. [6]

     Y. Wang, N.Y. Huang, J.Q. Shen, et al., J. Am. Chem. Soc. 140 (2018) 38–41. doi: 10.1021/jacs.7b10107

  7. [7]

     H. Zhong, R. Sa, H. Lv, et al., Adv. Funct. Mater. 30 (2020) 2002654. doi: 10.1002/adfm.202002654

  8. [8]

     M. Ding, H.L. Jiang, ACS Catal. 8 (2018) 3194–3201. doi: 10.1021/acscatal.7b03404

  9. [9]

     H. Rao, L.C. Schmidt, J. Bonin, M. Robert, Nature 548 (2017) 74–77. doi: 10.1038/nature23016

  10. [10]

      D. Mellmann, P. Sponholz, H. Junge, M. Beller, Chem. Soc. Rev. 45 (2016) 3954–3988. doi: 10.1039/C5CS00618J

  11. [11]

      P. Wang, R. Dong, S. Guo, et al., Nat. Sci. Rev. 7 (2020) 1459–1467. doi: 10.1093/nsr/nwaa112

  12. [12]

      Z. Guo, S. Cheng, C. Cometto, et al., J. Am. Chem. Soc. 138 (2016) 9413–9416. doi: 10.1021/jacs.6b06002

  13. [13]

      Q. Fu and C. Draxl, Phys. Rev. Lett. 122 (2019) 046101. doi: 10.1103/PhysRevLett.122.046101

  14. [14]

      T. Ouyang, H.H. Huang, J.W. Wang, D.C. Zhong, T.B. Lu, Angew. Chem. Int. Ed. 56 (2017) 738–743. doi: 10.1002/anie.201610607

  15. [15]

      V.S. Thoi, N. Kornienko, C.G. Margarit, P. Yang, C.J. Chang, J. Am. Chem. Soc. 135 (2013) 14413–14424. doi: 10.1021/ja4074003

  16. [16]

      T. Ishizuka, A. Watanabe, H. Kotani, et al., Inorg. Chem. 55 (2016) 1154–1164. doi: 10.1021/acs.inorgchem.5b02336

  17. [17]

      L. Zhang, X.X. Li, Z.L. Lang, et al., J. Am. Chem. Soc. 143 (2021) 3808–3816. doi: 10.1021/jacs.0c11450

  18. [18]

      N. Li, J. Liu, B.X. Dong, Y.Q. Lan, Angew. Chem. Int. Ed. 59 (2020) 20779–20793. doi: 10.1002/anie.202008054

  19. [19]

      L. Ma, Q. Wang, W.L. Man, et al., Angew. Chem. Int. Ed. 54 (2015) 5246–5249. doi: 10.1002/anie.201500507

  20. [20]

      L.Z. Dong, L. Zhang, J. Liu, et al., Angew. Chem. Int. Ed. 59 (2020) 2659–2663. doi: 10.1002/anie.201913284

  21. [21]

      Z. -. H. Yan, M.H. Du, J. Liu, et al., Nat. Commun. 9 (2018) 3353. doi: 10.1038/s41467-018-05659-7

  22. [22]

      D. Hong, Y. Tsukakoshi, H. Kotani, T. Ishizuka, T. Kojima, J. Am. Chem. Soc. 139 (2017) 6538–6541. doi: 10.1021/jacs.7b01956

  23. [23]

      N. Elgrishi, M.B. Chambers, X. Wang, M. Fontecave, Chem. Soc. Rev. 46 (2017) 761–796. doi: 10.1039/C5CS00391A

  24. [24]

      J.J. Liu, N. Li, J.W. Sun, et al., ACS Catal. 11 (2021) 4510–4519. doi: 10.1021/acscatal.0c04495

  25. [25]

      N. Bakharev, D. Bono, H.B. Brom, et al., Phys. Rev. Lett. 96 (2006) 117002. doi: 10.1103/PhysRevLett.96.117002

  26. [26]

      N. Zheng, X. Bu, P. Feng, J. Am. Chem. Soc. 124 (2002) 9688–9689. doi: 10.1021/ja020480p

  27. [27]

      H. Rao, J. Bonin and M. Robert, ChemSusChem. 10 (2017) 4447–4450. doi: 10.1002/cssc.201701467

  28. [28]

      H.X. Zhang, Q.L. Hong, J. Li, et al., Angew. Chem. Int. Ed. 58 (2019) 11752–11756. doi: 10.1002/anie.201905869

  29. [29]

      X.X. Li, L. Zhang, J. Liu, et al., JACS Au. 1 (2021) 1288–1295. doi: 10.1021/jacsau.1c00186

  30. [30]

      Y. Chen, S. Fan, B. Qiu, et al., ACS Appl. Mater. Interfaces 13 (2021) 24795–24803. doi: 10.1021/acsami.1c04155

  31. [31]

      B. Guo, Q. Li, J. Lin, et al., ACS Appl. Nano Mater. 4 (2021) 1849–1857. doi: 10.1021/acsanm.0c03224

  32. [32]

      Kashid, D.J. Patil, R.D. Mali, et al., Catal Letters. 151 (2021) 353–358. doi: 10.1007/s10562-020-03330-9

  33. [33]

      C. Liu, C.L. Ji, T. Zhou, X. Hong, M. Szostak, Angew. Chem. Int. Ed. 60 (2021) 10690–10699. doi: 10.1002/anie.202100949

  34. [34]

      J. Xu, Z. Ju, W. Zhang, et al., Angew. Chem. Int. Ed. 60 (2021) 8705–8709. doi: 10.1002/anie.202017041

  35. [35]

      T. Ouyang, H.J. Wang, H.H. Huang, et al., Angew. Chem. Int. Ed. 57 (2018) 16480–16485. doi: 10.1002/anie.201811010

  36. [36]

      S.M. Liu, Z. Zhang, X. Li, et al., Adv. Mater. Interfaces 5 (2018) 1801062. doi: 10.1002/admi.201801062

  37. [37]

      J.Z. Bloh, R. Dillert, D.W. Bahnemann, J. Phys. Chem. C 116 (2012) 25558–25562. doi: 10.1021/jp307313z

  38. [38]

      X. Feng, F. Pan, H. Zhao, et al., Appl. Catal. B 238 (2018) 274–283. doi: 10.1016/j.apcatb.2018.07.027

  39. [39]

      Z. Jiang, J. Liu, M. Gao, et al., Adv. Mater. 29 (2017) 1603369. doi: 10.1002/adma.201603369

  40. [40]

      B. Tang, X. Yang, Z. Kang, L. Feng, Appl. Catal. B 278 (2020) 119281. doi: 10.1016/j.apcatb.2020.119281

  41. [41]

      B. Liu, Y.F. Zhao, H.Q. Peng, et al., Adv. Mater. 29 (2017) 1606521. doi: 10.1002/adma.201606521

  42. [42]

      Z. Li, H. Dong, Y. Zhang, et al., J. Phys. Chem. C 112 (2008) 16046–16051. doi: 10.1021/jp805475a

  43. [43]

      B. Zhang, J. Zhang, F. Zhang, et al., Adv. Funct. Mater. 30 (2020) 1906194. doi: 10.1002/adfm.201906194

  44. [44]

      L. Feng, K.Y. Wang, E. Joseph, H.C. Zhou, Trends Chem. 2 (2020) 555–568. doi: 10.1016/j.trechm.2020.01.003

  45. [45]

      S.N. Habisreutinger, L. Schmidt-Mende, J.K. Stolarczyk, Angew. Chem. Int. Ed. 52 (2013) 7372–7408. doi: 10.1002/anie.201207199

  46. [46]

      J.P. Beale, J.A. Cunningham, D.J. Philips, Inorg. Chim. Acta. 33 (1979) 113–118. doi: 10.1016/S0020-1693(00)89463-3

  47. [47]

      N. Pilkington, R. Robson, Aust. J. Chem. 23 (1970) 2225–2236.

  48. [48]

      X. Chang, T. Wang, J. Gong, Energy Environ. Sci. 9 (2016) 2177–2196. doi: 10.1039/C6EE00383D

  49. [49]

      X.K. Wang, J. Liu, L. Zhang, et al., ACS Catal. 9 (2019) 1726–1732. doi: 10.1021/acscatal.8b04887

  50. [50]

      Y. Fu, D. Sun, Y. Chen, et al., Angew. Chem. Int. Ed. 51 (2012) 3364–3367. doi: 10.1002/anie.201108357

  51. [51]

      N. Li, J.J. Liu, J.W. Sun, et al., Green Chem. 22 (2020) 5325–5332. doi: 10.1039/D0GC01497D

• 

  Figure 1  Structures of the molecular complexes. (a) mononuclear Ni-L¹; (b) mononuclear Cu-L¹; (c) dinuclear Ni₂-L²; (d) dinuclear Cu₂-L²; (e) dinuclear CuNi-L². All hydrogen atoms are omitted for clarity. Cu, yellow; Ni, green; O, red; N, blue; C, gray.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) High resolution Ni 2p XPS spectra of Ni-L¹, Ni₂-L² and CuNi-L² in survey scan. (b) Cyclic voltammogram of CuNi-L² in acetonitrile containing 0.1 mol/L Bu₄NPF₆ at a scan rate of 50 mV/s. (c) UV–vis absorption spectra of Ni-L¹, Cu-L¹, Ni₂-L², Cu₂-L² and CuNi-L² in the solution state (chloroform). (d) Band structure for Ni-L¹, Ni₂-L² and CuNi-L².

  下载: 全尺寸图片 幻灯片
  

  Figure 3  The photocatalytic CO₂RR was carried out under the conditions of a 30 mL CO₂-saturated CH₃CN/H₂O solution (v/v = 4:1) at 25 ℃ with the same concentration of Ni^II (0.8 µmol/L), TIPA (0.8 mol/L) and [Ru(bpy)₃]Cl₂·6H₂O (0.85 mmol/L) under visible light irradiation (λ ≥ 420 nm, irradiation area, 7.06 cm²). (a) Photocatalytic production of CO and H₂ catalyzed by Ni-L¹ (0.8 µmol/L), Ni₂-L² (0.4 µmol/L) and CuNi-L² (0.8 µmol/L). (b) Mounts of CO and H₂ produced as a function of the visible-light irradiation time over CuNi-L² (0.8 µmol/L). (c) Durability test for CuNi-L² (0.8 µmol/L). (d) The mass spectra of ¹³CO recorded under a ¹³CO₂ atmosphere.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) The free-energy profile for the CO₂RR pathway. (b) Charge density difference plot for the first excited state of CuNi-L² (left) and Ni₂-L² (right). (c) 2D display of charge density of CuNi-L² (left) and Ni₂-L² (right) systems. (d) Proposed photocatalytic mechanism of the molecular metal compounds for the CO₂ to CO conversion.

  下载: 全尺寸图片 幻灯片
•